• Keine Ergebnisse gefunden

Radiation, biological diversity and host-parasite interactions in wild roses and insects

N/A
N/A
Protected

Academic year: 2021

Aktie "Radiation, biological diversity and host-parasite interactions in wild roses and insects"

Copied!
121
0
0

Wird geladen.... (Jetzt Volltext ansehen)

Volltext

(1)

Radiation, biological diversity and

host-parasite interactions in wild roses and

insects

Dissertation

zur Erlangung des Doktorgrades der Naturwissenschaften

(Dr. rer. nat.)

dem

Fachbereich Biologie

der Phillipps-Universität Marburg

vorgelegt von

Annette Kohnen

aus Darmstadt

(2)

Vom Fachbereich Biologie der Philipps-Universität Marburg als Dissertation angenommen am: 22.10.2009

Erstguachter: Prof. Dr. Roland Brandl Zweitgutachter: Prof. Dr. Volker Wissemann

(3)
(4)

1. Einleitung ... 1

1.1 Radiationen und ihre Auswirkungen...1

1.2 Geographische Diversitätsmuster ...2

1.3 Das Modelsystem der Wildrosen...2

1.4 Gallenbildung durch Diplolepis rosae...3

1.5 Die Hagebuttenfruchtfliegen Rhagoletis alternata und Carpomya schineri ...4

1.6 Fragestellungen und Ergebnisse...5

1.7 Schlussfolgerung...8

2. Radiation, biological diversity and host-parasite interactions in wildroses,

rust fungi and insects. ... 9

2.1 Introduction: Radiation, biodiversity and host-parasite interaction in the Rosa-system10 2.2 Dog rosese are allopolylpoids: Genetic constitution of section Caninae ...11

2.3 Character inheritance in the heterogamous system of dog roses ...13

2.4 Glandular trichomes matter: Rust fungi on Rosa ...15

2.5 Evolution and diversity of plant-pathogen-insect foodwebs on dog roses...15

2.6 How are the differences between the three closely related dog rose species translated into higher trophic levels? ...25

2.7 Conclusion...27

3. Cynipid gall-parasitoid interactions, comparing three dog rose species along

a geographical gradient... 29

4. No host-associated differentiation in the gall wasp Diplolepis rosae

(Hymenoptera: Cynipidae) on three dog rose species ... 43

(5)

5. Comparing geographical structures of one cynipid gall wasp with two

specialised parasitoids in Europe ... 56

6. No genetic differentiation in the rose-infesting fruit flies Rhagoletis alternata

and Carpomya schineri (Diptera: Tephritidae) across central Europe ... 77

7. Summary ... 84

8. Referenzen ... 88

9. Appendix ... 104

Erklärung zu eigenen Beiträgen und Veröffentlichungen... 109

Erklärung... 111

Danksagung ... 112

Lebenslauf... 114

(6)
(7)

1. Einleitung

_____________________________________________________________________________________________________

1

1. Einleitung

1.1 Radiationen und ihre Auswirkungen

Aus einem komplizierten Geflecht gegenseitiger Abhängigkeit verschiedenster Organismen haben sich zum Teil hochspezialisierte Interaktionen entwickelt. Dazu zählen Symbiosen, Mutualismen, parasitische und prädatorische Lebensweisen auf mehreren trophischen Ebenen. Gegenseitige Abhängigkeit kann zu einer Koevolution zwischen den interagierenden Partnern und damit zu Differenzierungen innerhalb von Arten führen. So wird angenommen, dass die hohe Artenvielfalt (Biodiversität) der heutigen Blütenpflanzen (Angiospermen) und ihrer abhängigen Insektenfauna durch Interaktionen beider Partner entstanden ist (z.B. Ehrlich & Raven 1964).

Für die Entstehung hoher Biodiversität durch Artbildung (Speziation) sind evolutionäre Prozesse, wie räumliche oder zeitliche Trennungen, notwendig. Räumliche Trennungen können sowohl durch geographische Barrieren, als auch durch Wirtswechsel verursacht werden. In Europa sind viele Arten durch den Klimawandel der letzten Eiszeiten geprägt (Hewitt 1996, Taberlet et al. 1998, Hewitt 2000). Phänotypische Unterschiede zwischen Wirtspflanzen können z.B. durch unterschiedliche Fruchtungs- oder Blühzeitpunkte zu zeitlichen Trennungen von sich adaptierenden Arten führen (Bush 1969, Drès & Mallet 2002). Durch Anpassungen an Umweltbedingungen und das Ausnutzen neuer ökologischer Nischen kann es zu einer Aufteilung einer Art in mehrere höher spezialisierte Arten kommen, sogenannten Radiationen. Bekannte Beispiele für Radiationen sind die Darwinfinken auf Galapagos (Grant 1986), die Kleidervögel auf Hawaii (Wagner & Funk 1995) und die Buntbarsche (Cichliden) der ostafrikanischen Seen (Fryer & Iles 1972). Aber auch viele Pflanzenarten haben während der letzten Eiszeiten Radiationen durchlaufen; ein Beispiel in Europa sind die Wildrosen der Sektion Caninae (Wissemann 2005, Ritz et al. 2005b).

Besonders wichtig für Speziationsprozesse scheint vor allem die genetische Diversität zu sein. Durch ein Radiationsereignis entsteht aus einer Art in relativ kurzer Zeit eine Vielzahl genetischer und phänotypischer Variabilität. Über die Auswirkungen genetischer Variabilität auf Populationsstrukturen wurde in den vergangenen Jahren viel geforscht (z. B. Barratt et al. 1999, Barrowclough et al. 2005, Zenger et al. 2005). Aufgrund der genetischen Diversität einer Art wurden Überlebens- und Aussterbeszenarien prognostiziert, Ausbreitungswege rekonstruiert und Anpassungsprozesse postuliert. Darüber hinaus stellt sich die Frage, wie sich eine relativ junge, hohe Diversität, entstanden durch eine Radiation, auf abhängige

(8)

Arten wie Herbivore, Räuber oder Parasiten auswirkt. Beginnen sich die abhängigen Arten an die neue Formenvielfalt zu adaptieren, spezialisieren sie sich auf einzelne Arten und führt dies in der Folge ebenfalls zu einer Radiation und damit zu hoher Diversität der abhängigen Arten?

1.2 Geographische Diversitätsmuster

Auch phylogeographische Muster und räumliche Strukturen abhängiger Arten können durch ihre Wirte geprägt sein (z.B. Nieberding et al. 2004). Diversitätsmuster vieler Tier- und Pflanzenarten in Europa sind stark durch die Temperaturschwankungen des Pleistozäns geprägt (Hewitt 1996). Durch die Bedeckung Europas mit Gletschern waren viele Arten gezwungen sich in wärmere, eisfreie Gebiete, die Refugialgebiete, zurückzuziehen. Die mediterranen Regionen der iberischen und der Apennin-Halbinsel, sowie der Balkan bildeten solche Rückzugsgebiete (z.B. Hewitt 1996, Oshida et al. 2005). Nach dem Ende der Kaltzeiten breiteten sich viele Arten über Rückwanderungsrouten wieder nach Mitteleuropa aus (Taberlet et al. 1998). Aufgrund der heutigen Diversitätsmuster innerhalb der Arten kann man Rückschlüsse auf ihre Refugialgebiete und ihre Rückwanderwege ziehen.

Zu erwarten wäre, dass interagierende Arten gleiche Umweltbedingungen und eine gemeinsame Historie teilen, zudem sollten Refugialgebiete und Rückwanderwege zusätzlich von der jeweils anderen Art abhängen. Zu erwarten wären gleiche oder zumindest ähnliche geographische Muster. Für mehrere Wirt-Parasiten-Systeme wurden solche Strukturen schon öfters gezeigt (Funk et al. 2000, Nieberding et al. 2004, LaJeunesse et al. 2004). Nun stellt sich die Frage, ob das auch auf Wirte und ihre Parasitoiden zutrifft, die ähnlich wie Parasiten auf ihren Wirt spezialisiert sind.

1.3 Das Modelsystem der Wildrosen

Für beide Fragestellung eignet sich das System der Wildrosen, speziell der Sektion Caninae (Gattung: Rosa, Sektion Caninae (DC.) Ser.) besonders, da diese Rosen zeitgeschichtlich betrachtet erst vor kurzem eine Radiation durchlaufen haben und eine hohe Diversität aufweisen (Wissemann 2005, Ritz et al. 2005b). Die Sektion Caninae ist auf Hybridisierungs-Ereignisse während und nach den Eiszeiten (Pleistozän) zurückzuführen (Ritz et al. 2005b). Ein Hinweis auf eine Entstehung der Sektion im Zuge der pleistozänen Trennungs- und Einwanderungsereignisse ist ihr allopolyploider Chromosomensatz (Gustafsson & Hakansson, 1942). Sie verfügen über einen fünf-fachen Chromosomensatz, der in einem komplizierten Meiose-Verfahren (Reifeteilung) weitergegeben wird. Dabei stammen 4/5 des Genoms von mütterlicher und nur 1/5 von väterlicher Seite (Ritz & Wissemann 2003). Man unterscheidet fast 200 Rosenarten, die auch heute noch in der Lage sind zu hybridisieren.

(9)

1. Einleitung

_____________________________________________________________________________________________________

3

Abb.1.1. Lebenszyklen der rosespezifischen Hagebuttenfruchtfliege (Rh. alternata) und der Rosengallwepe (D. rosae). (Verschiedene Quellen © der Fotos u.a. Randolph, 2005; S.Rösner, A. Vaupel).

Untersucht werden sollte, inwieweit sich diese Diversität auf rosenspezifische Insekten, die Hagebuttenfruchtfliegen (Rhagoletis alternata Fall. 1820 und Carpomya schineri Loew. 1856) und die Rosengallwespe (Diplolepis rosae L.) mit ihren spezifischen Parasitoiden auswirkt. Untersucht wurde zum einen eine Differenzierung zwischen verschiedenen Wirtspflanzen, zum anderen eine geographische Differenzierung. Um eine Anpassung an verschiedene Rosenarten zu überprüfen (Konzept der Wirtsrassen), wurden drei Rosenarten ausgewählt (Rosa canina L., R. corymbifera Borkh. und R. rubiginosa L.). Alle drei Arten sind in Europa weit verbreitet und kommen häufig in größerer Anzahl in denselben Gebieten vor (Timmermann & Müller 1994). Trotz ihrer nahen Verwandtschaft unterscheiden sie sich in mehreren, für beide Insektenarten relevanten, Merkmalen (z. B. Fruchtungszeitpunkt, Behaarung, sekundäre Inhaltsstoffe; Wissemann et al. 2006, Timmermann 1998).

1.4 Gallenbildung durch Diplolepis rosae

Für multitrophische Untersuchungen sind die Gallen der Rosengallwespe D. rosae sehr gut geeignet, da in ihnen Parasitoide und Hyperparasitoide mehrerer trophischer Stufen zu finden sind. Zudem sollten sich Präferenzen für verschiedene Rosenarten durch das parthenogenetische Fortpflanzungssystem, induziert durch das Bakterium Wolbachia, schneller abbilden.

(10)

Weibchen der Rosengallwespe fliegen im Juni/Juli und legen ihre Eier in Sprosse der Wildrosen (Schröder 1967). Durch das Ablegen der Eier und das Schlüpfen der Larven wird die Bildung einer Pflanzengalle induziert (Abb.1.1). Das Gewebe der Rose beginnt zu wuchern und wächst zu einer ballförmigen Kugel mit filzigen Auswüchsen heran, die den Larven sowohl Nahrung durch spezielles Nährgewebe, als auch Schutz vor Fraßfeinden bietet (Randolph 2005). Im Herbst, wenn die Larven ausgewachsen sind, stirbt das Rosengewebe ab. Die Larven überdauern den Winter in der Galle. Im Frühling, ausgelöst durch wärmere Temperaturen, beginnen sich die Larven zu verpuppen und schlüpfen wenige Wochen später zwischen Mai und Juli. Trotz des Schutzes, den die Galle den Larven bietet, hat sich eine Reihe von anderen Insekten auf Diplolepis-Gallen spezialisiert. Neben dem Inquilin Periclistus brandtii Ratzeburg (Hym., Cynipidae) sind dies neun häufige Parasitoide (Askew 1960, Stille 1984). Während der Inquilin nur das Gallengewebe für seine eigenen Gallenkammern nutzt, ernähren sich die Parasitoiden-Larven von den Gallwespen-Larven. Die beiden häufigsten Parasitoide in den Gallen von D. rosae sind die Schlupfwespe

Orthoplema mediator Thunb. (Hym. Ichneumonidae) und die Erzwespe Glyphomerus stigma

Fabr. 1793 (Hym. Torymidae).

Induziert durch das Bakterium Wolbachia ist D. rosae in der Lage sich parthenogenetisch fortzupflanzen (Schilthuizen & Stouthamer 1998). Infizierte Weibchen können sich nicht mehr mit Männchen ihrer Art verpaaren. Sie legen unbefruchtete Eier, die sich mit Hilfe des Bakteriums zu diploiden Nachkommen entwickeln. Alle Nachkommen werden weiblich, was zu einer Abweichung vom 1:1 Verhältnis zwischen Männchen und Weibchen führt (McCallan 1940). Männchen werden immer seltener, während die Population fast ausschließlich aus Weibchen besteht. Es findet kaum noch genetischer Austausch statt, der bei anderen Tieren durch die Paarung und Rekombination sichergestellt wird.

1.5 Die Hagebuttenfruchtfliegen Rhagoletis alternata und Carpomya schineri

Die Hagebuttenfruchtfliegen Rh. alternata und C. schineri wurden als weitere rosenspezifische Arten ausgewählt. Erstere zählt zur selben Gattung wie die Apfelfruchtfliege (Rhagoletis pomonella; Walsh), dem Modellorganismus für Wirtsrassenbildung (Bush 1969). Beide Arten kommen in Südeuropa sympatrisch vor. Während Rh. alternata im gesamten paläarktischen Raum verbreitet ist (Kandybina 1977), ist C. schineri auf die Südpaläarktis beschränkt (White & Elson-Harris 1992). Beide Fruchtfliegen sind eng an ihre Wirtspflanze gebunden (Abb.1.1). Weibchen legen im Juni jeweils ein Ei in die noch grüne Hagebutte und markieren sie mit einem Pheromon (Bauer 1986). Die Larven ernähren sich ausschließlich vom Fruchtfleisch der Hagebutte und beschädigen die Samen nicht. Die Beziehung zwischen Fruchtfliegen und Wirtspflanze wird daher als „nicht-interaktiv“ bezeichnet.

(11)

1. Einleitung

_____________________________________________________________________________________________________

5 Wirtspflanze (Bauer 1998). Die vollentwickelten Larven von Rh. alternata verlassen im Oktober, die von C. schineri bereits im August, die reife Hagebutte, um sich über den Winter in der Erde zu verpuppen (Hendel 1927; Bauer 1986). Trotz geringer Unterschiede zwischen Fliegen auf verschiedenen Rosenarten, konnten bisher für Rh. alternata auf der Basis von Allozymen keine Wirtsrassen nachgewiesen werden (Leclaire & Brandl 1994, Vaupel et al. 2007). Ebenfalls wurde mit Allozymen nur eine geringe Differenzierung von Rh. alternata zwischen Standorten innerhalb Deutschlands und der Schweiz nachgewiesen (Vaupel et al. 2007). Selbst die Alpen scheinen keine Barriere für die Rosenfruchtfliege zu bilden.

1.6 Fragestellungen und Ergebnisse

1.6.1: Gibt es Unterschiede in der Befallsrate von Diplolepis rosae auf verschiedenen Rosenarten? Ist die Parasitierungsrate auf diesen Rosenarten unterschiedlich? (Kapitel 3)

Um Wirtspräferenzen der Rosengallwespe D. rosae zu ermitteln wurden auf drei verschiedenen Rosenarten (R. canina, R. corymbifera und R. rubiginosa) an 17 Standorten in Deutschland die Anzahl der Gallen aufgenommen. Zusätzlich wurden Höhe, Blattdichte, Hagebuttendichte und Anzahl der Sprosse an je 5 Büschen ermittelt. An acht von diesen Standorten wurden alle Gallen abgesammelt, aufbewahrt und alle schlüpfenden Insekten (Gallwespen und Parasitoide) bestimmt und gezählt, insgesamt 6175 Individuen aus 388 Gallen.

Mit Hilfe eines Generalisierten Linearen Modells (GLM) wurde der Einfluss der Rosenart und des Standortes auf die Anzahl der Gallen pro Busch untersucht. Es zeigte sich, dass die Befallsrate mit Gallen stärker von der Form des Busches abhängt, als von der Rosenart. Büsche mit vielen austreibenden Ästen werden häufiger befallen, da sie den Gallwespen mehr Möglichkeiten zur Induktion von Gallen geben. Die Wuchsform eines Busches unterscheidet sich jedoch zwischen den Rosenarten. Im Mittel zeigt R. rubiginosa einen buschigeren und höheren Wuchs, wird somit im Mittel häufiger befallen als die anderen beiden Rosenarten. Die Größe der gebildeten Gallen ist jedoch unabhängig von der Rosenart.

In einer zweiten Untersuchung wurde der Einfluss der Rosenart auf die Parasitoide getestet. Ebenfalls mit einem GLM konnte gezeigt werden, dass die Parasitierungsrate nicht von der Rosenart, sondern vom Standort und vom Gallvolumen abhängt. Es besteht allerdings eine starke Interaktion zwischen der Rosenart und dem Standort. Die Zusammensetzung der Parasitoiden-Gemeinschaft ist vom Gallvolumen, aber auch vom Standort und der Rosenart abhängig.

(12)

Daraus kann geschlossen werden, dass sich Interaktionen multitrophischer Artengemeinschaften nicht zwischen Rosenarten unterscheiden. Wahrscheinlich unterscheiden sich die Wildrosen durch fortgesetzte Hybridisierungen nicht genug, um Präferenzen für einzelne Rosenarten auszubilden. Trotzdem zeigt die Wirtspflanze einen wichtigen Einfluss, der sich durch verschiedene Umweltbedingungen an verschiedenen Standorten auswirkt.

1.6.2: Unterscheiden sich Gallwespen auf verschiedenen Rosenarten genetisch voneinander? (Kapitel 4)

Im Herbst 2006 und 2007 wurden an fünf Standorten in Deutschland alle verfügbaren Gallen der drei Rosenarten (R. canina, R. corymbifera, R. rubiginosa) gesammelt. Die Gallen wurden in Plastikboxen mit Gazeverschluss (Luftaustausch) im Freien gelagert, um eine natürliche Entwicklung zu gewährleisten. Ab Mai des darauffolgenden Jahres schlüpften sowohl die adulten Gallwespen als auch die Parasitoide. Die geschlüpften Tiere wurden ausgezählt, sortiert und konserviert. Pro Standort wurden von jeder Rosenart maximal 15 (insgesamt 149) Individuen untersucht. Sie wurden auf ihren Befall mit dem parthenogenese-induzierenden Bakterium Wolbachia überprüft und mit fünf Primerkombinationen der populationsgenetischen Methode der AFLPs (Vos et al. 1995) analysiert. Die Befallsrate mit

Wolbachia Bakterien betrug fast 100%. Dagegen zeigten die 106 polymorphen AFLP Marker

weder genetische Differenzierungen zwischen Wirtspflanzen noch zwischen geographischen Standorten. Dies deutet auf eine gute Ausbreitungsfähigkeit u.a. durch Windverdriftung, sowie möglicherweise mehrere Infektions-Ereignisse mit erfolgreichen Wolbachia-Stämmen hin, die sich in der ganzen Population durchsetzen konnten. Die Wahl der Wirtspflanze scheint hierbei keine genetische Austauschbarriere zu sein, da es keine Spezialisierung von Gallwespen auf unterschiedliche Wirtspflanzen zu geben scheint. Ein Grund hierfür könnte die fortdauernde Hybridisierung zwischen verschiedenen Wildrosenarten sein.

1.6.3: Sind die Diversitätsmuster von Diplolepis rosae und seinen häufigsten Parasitoiden

Orthopelma mediator und Glyphomerus stigma europaweit ähnlich? (Kapitel 5)

Alle drei Insektenarten sind in ihren Lebenszyklen eng miteinander verbunden und beide Parasitoide befallen ausschließlich Arten der Gattung Diplolepis, die wiederum alle auf Wildrosenarten angewiesen sind. Daher wurden ähnliche genetische Strukturen aufgrund ähnlicher Historien innerhalb Europas erwartet.

Gesammelte Gallen aus verschiedenen Ländern Europas wurden im Freien in Plastikboxen gelagert, so dass sich alle Larven entwickeln und schlüpfen konnten. Sodann wurden die Insekten aussortiert, bestimmt und in Ethanol konserviert. Von 79 D. rosae Individuen aus 17 europäischen Ländern, 56 O. mediator und 28 G. stigma Individuen wurde der Interne

(13)

1. Einleitung

_____________________________________________________________________________________________________

7 Transkribierte Spacer 2 (ITS 2) aus dem Genom und die Cytochrom Oxidase I (COI) vom Mitochondrium sequenziert. Alle Individuen wurden auf ihren Befall mit Wolbachia Bakterien untersucht. Es zeigte sich, dass die Diversitätsmuster und genetischen Strukturen der drei Insektenarten innerhalb Europas sehr unterschiedlich sind. Die Gallwespe, D. rosae, zeigt eine geringe Variabilität und einen hohen Grad an Durchmischung, was auf eine gute Ausbreitungsfähigkeit hinweist. Sie ist fast komplett mit Wolbachia Bakterien befallen. In Südeuropa ist die genetische Diversität höher als in Zentraleuropa, was auf südliche Refugialgebiete hindeutet. Ganz anders ist das Muster des Parasitoiden O. mediator. Diese Insektenart zeigt eine klassische Ost-West Trennung zweier Linien in Europa mit einer Suturzone in Frankreich und Deutschland. Beide Linien sind deutlich differenziert, erklärbar durch zwei Refugialgebiete möglicherweise in Spanien und Osteuropa und ihre entsprechenden Rückwanderungsrouten. Der zweite Parasitoid, G. stigma, zeigt eine extrem hohe Variabilität, die keine Rückschlüsse auf räumliche Strukturen zuläßt und ist überhaupt nicht mit Wolbachia befallen. Die extremen Unterschiede in den Populationsstrukturen und deren Entkopplung ist unerwartet und nur durch ein Ausweichen auf andere Wirtsarten zu erklären.

1.6.4: Wie groß ist die Diversität von Rosenfruchtfliegen europaweit? (Kapitel 6)

Larven von Rh. alternata und C. schineri wurden in Deutschland, sowie den angrenzenden europäischen Ländern gesammelt. Befallene Hagebutten wurden in perforierten Plastikbeuteln bei ca. 15°C gelagert bis die Larven die Hagebutten verließen, um sich zu verpuppen. Bis zur Verwertung im Labor wurden die Puppen in 90% Ethanol gelagert. Es wurden Rh. alternata Larven von 12 Standorten und C. schineri Larven von fünf Standorten untersucht. Drei in der Regel sehr variable Gene (Cytochrom oxidase I und II, Cytochrom b), die sich für populationsgenetische Untersuchungen eignen (Simon et al. 1994, z.B. Lunt et al. 1996, Rokas et al. 2002, Simon et al. 2006), wurden sequenziert. Diese drei mitochondriellen Gene ergeben einen Sequenzabschnitt von 1720 Basenpaaren (bp). Es hat sich gezeigt, dass alle untersuchten Rh. alternata Individuen exakt die gleiche Basenpaarfrequenz aufweisen, also keinerlei Diversität gefunden werden konnte. Die Diversität von C. schineri war ebenfalls sehr gering, es konnten nur zwei Haplotypen gefunden werden. Ein Vergleich mit 61 anderen Insektenarten zeigt, dass dies eine außergewöhnlich geringe Variabilität ist. Dies ist ein sehr erstaunliches Ergebnis und nur zum Teil durch die gute Verbreitungsfähigkeit und hohen Populationsdichten der beiden Fruchtfliegen zu erklären.

(14)

1.7 Schlussfolgerung

Die grundlegende Fragestellung dieser Untersuchung war, in welchem Grad die hohe Variabilität der Wildrosen in den abhängigen Arten der nächst höheren trophischen Stufen wiederzufinden ist. Für keine der beide rosenspezifischen Insektenarten, weder für die Gallwespe D. rosae, noch für die Hagebuttenfruchtfliege Rh. alternata (Leclaire & Brandl 1994, Klinge 2005, Vaupel et al. 2007), konnten bisher Wirtspräferenzen oder gar Adaptationen gefunden werden. Ein ähnliches Ergebnis wurde von Ritz et al. (2005a) für Pilze der Gattung Phragmidium gefunden, sie zeigen ebenfalls keine Unterschiede auf verschiedenen Rosenarten. Auch die dritte trophische Ebene, die Parasitoide, wurden durch die Wirtspflanze kaum beeinflusst. Viel wichtiger scheint für die Artenzusammensetzung die Gallengröße zu sein. Zu erklären ist der Mangel an Adaptation durch die fortdauernde Hybridisierung der heutigen Rosenarten, die zu einem ständigen Genfluss zwischen den Arten führt und damit scharfe Grenzen zwischen den Arten verwischt. Hybride zwischen Wildrosen erben ihre phänotypischen Ausprägungen zum Großteil von der Mutterpflanze (Wissemann et al. 2006), jedoch sind alle Zwischenstufen durch fortlaufende Hybridisierung möglich. Das kann für die abhängigen Insekten einen Wechsel von einer Art auf eine andere Art ohne Schwierigkeiten ermöglichen.

Auch geographisch zeigt sich kein übereinstimmendes Bild in der genetischen Struktur sowohl zwischen den rosenspezifischen Insektenarten, als auch zwischen ihnen und ihren Parasitoiden. Damit sind keine Rückschlüsse auf eine Historie der Wirtspflanzen zu ziehen. Die Entkopplung und völlige Abweichung ist vermutlich auf unterschiedliche Ausweichmöglichkeiten in der Wirtswahl zurückzuführen. Einerseits können die rosenspezifischen Insektenarten zwischen verschiedenen Rosenarten, teilweise sogar Rosen anderer Sektionen, wechseln, andererseits sind auch die Parasitoide in der Lage andere Wirte der Gattung Diplolepis zu befallen, die wiederum ebenfalls andere Rosensektionen befallen können. Zudem haben alle untersuchten Insektenarten gute Ausbreitungsfähigkeiten aufgrund ihrer geringen Größe und damit guten Verdriftbarkeit durch Wind. Rosen, auch Hundsrosen wurden und werden immer noch in ihrer Ausbreitung stark vom Menschen beeinflusst. Sie werden u. a. in Gärten, Parkanlagen, als Hecken und entlang von Wegrändern und Eisenbahnlinien angepflanzt. Mit den Pflanzen können auch die Insekten leicht verbreitet werden. Das erklärt den hohen Grad an genetischem Austausch und die geringe Differenzierung zwischen den Standorten bei den meisten der untersuchten Insekten. Im System der Wildrosen und ihrer abhängigen Arten finden sich demnach weder wirtsspezifische, noch einheitliche geographische Differenzierungen in den höheren trophischen Ebenen wieder.

(15)

2. Radiation, biological diversity and host-parasite

interactions in wildroses, rust fungi and insects.

Annette

K

OHNEN

, Roland

B

RANDL

, Roman

F

RICKE

,

Friederike

G

ALLENMÜLLER

,

Katrin

K

LINGE

,

Ines

K

ÖHNEN

,

Wolfgang

M

AIER

,

Franz

O

BERWINKLER

,

Christiane

R

ITZ

,

Thomas

S

PECK

,

Günter

T

HEISSEN

,

Teja

T

SCHARNTKE

,

Andrea

V

AUPEL

,

Volker

W

ISSEMANN

A

BSTRACT

One of the major tasks in evolutionary ecology is to explain how interspecific interactions influence the dynamics of evolutionary processes and enable radiation and genesis of biological diversity. The bewildering diversity of dog roses is generated by a heterogamous reproductive system. Genetic distance between rose taxa was analysed as base line for the explanation of subsequent radiation of the two host dependent parasite groups, rust fungi and insects. We investigated the interaction between each host-parasite system and between the parasite groups. We learned that the functional diploidy at the meiotic level is not reflected at the phenotypic level in dog roses. The phytophagous insect community shows only minor differences in composition on different rose species. These invertebrates seems not negatively affected by glandular trichomes, but for the rust fungi Phragmidium „glandular trichomes matter“, because they are negatively correlated with the infection. The abundance of two rose specialists the rose hip fly Rhagoletis alternata Fall. and the rose gall wasp Diplolepis rosae L. differed on rose species, but Rh. alternata showed neither any genetic differentiation on host species, nor geographical differentiation. As a basic result we detected that genetic diversity of dog roses is not translated into a hostspecific radiation of the parasites. We assume that intensive reticulate evolution of dog roses prevents cospeciation.

(16)

2.1 Introduction: Radiation, biodiversity and host-parasite interaction in the

Rosa

-system

A key topic in evolutionary ecology is to explain how interspecific interactions influence the dynamics of evolutionary processes and enable radiation and unfolding of biological diversity. In host-parasite interacting systems the most important questions is: How does the radiation and diversity of the hosts translate into the radiation and diversity of the parasites and what is the role of parasite interactions?

Such analyses are extremely rare due to the complexity of these systems (see e.g. Clay 1989, Pirozynski & Haksworth 1989). In this study we investigate the host-parasite net of dog roses (Rosaceae, Rosoideae, Rosa L., sect. Caninae (DC.) Ser.), rust fungi (Phragmidium), and insects. By analysing the genetic distance, variability and phylogenetic relationships between rose taxa, we determine a base line for the explanation of the subsequent radi-ation processes of two host dependent parasite groups, rust fungi and insects (Fig. 2.1). Understanding of the dog roses’ radiation process enables us to unravel the levels of interaction (co-evolution, co-speci-ation, individualistic interaction) on which rust fungi and insects act. Dog roses are thought to have evolved during the Pliocene (5.3 -1.8 Mya) as a result of a single event, into which the peculiar mode of Canina-meiosis developed and then colonised Central Europe by a very fast and explosive radiation since the Pleistocene and Holocene (Zielinski 1985). Wissemann (2000a) and the study by Ritz et al. (2005b) showed that dog roses are permanent allopolyploids

Fig. 2.1. A. lower surface of R. corymbifera (hairy), B. R. rubiginosa (glandular), C. R. canina (gla-brous, eglandular). D. Phragmidium mucronatum, E. Phragmidium tuberculatum, F. rust infection on a hip of R. canina, G. Teliospores, H. Rhagoletis alterna infected hip of R. canina, I. Flower diversity in Rosa, J. Hip diversity

(17)

2. Radiation, diversity and host-parasite interaction

_____________________________________________________________________________________________________

11 arisen by multiple hybridisation events. The high genetic variability due to allopolyploidy and great homology between the different chromosome sets enabling interfertility between any dog rose species are the reasons for the morphological variation and the existence of numerous local forms.

Subsequent to the Rosa radiation, numerous pathogens interacted with their hosts. At present both, the analysed species of rusts and insects are found on any of the investigated dog rose species. Nevertheless, little is known about the genetic diversity of the two main (and commercially important) rust fungi on roses, Phragmidium mucronatum (Pers.) Schltdl. and Phragmidium tuberculatum J. Müller. We do not know anything about the radiation process of subspecific taxa (races, strains) and we do not know the level on which the fungi became host specific. The same is the situation in the numerous rose specific insect species. One group are the gall-forming insects, which are highly susceptible to plant resistance and are adapted to specific resources in most cases.

2.2 Dog rosese are allopolylpoids: Genetic constitution of the section Caninae

The genus Rosa consists of about 200 species following the classification by Wissemann and Ritz (2005). Based on morphological characters four subgenera are recognized:

Hulthemia (Dumort.) Focke, Platyrhodon (Hurst) Rehder, Hesperhodos Cockerell, and Rosa.

The subgenus Rosa comprises ten sections with more than 150 species of various ploidy-levels, distributed mainly in the temperate regions of the northern hemisphere. Studies involving excessive cloning of nuclear genes (ribosomal spacers and low copy genes) revealed that hybridisation played an important role in the evolution of polyploid rose taxa (Wissemann 2000a, Ritz et al. 2005b, Joly & Bruneau 2006, Joly et al. 2006). Within subgenus Rosa, members of the polyploid section Caninae are of particular interest, not only because of their unique meiotic behaviour but also for the readiness, with which members of the section can hybridize (Feuerhahn & Spethmann 1995, Wissemann & Hellwig 1997, Reichert 1998, Nybom et al. 2004, 2006, Werlemark & Nybom 2001, Werlemark et al. 1999). Early studies analysing nrITS-1 sequences revealed the existence of non-concerted evolution of the nrITS-region and thus confirmed the allopolyploid constitution of dog roses (Eigner & Wissemann 1999, Wissemann 1999, 2000b, 2002). Based on these studies we showed, that the tetra-, penta- or hexaploid dog roses arose by multiple hybridisations across the genus Rosa (Fig. 2.2, Ritz et al. 2005b).

Dog roses are cytologically characterised by a specific meiosis (Täckholm 1920, 1922, Klášterská 1969, 1971, Klášterská & Natarajan 1974, Roberts 1975). This meiosis leads to

(18)

heterogamous reproduction with (in the case of pentaploid roses, 2n=5x=35) haploid pollen grains (n=1x=7) and tetraploid egg cells (n=4x=28; reviewed in Wissemann & Ritz 2007).

By microsatellite analysis (Ritz & Wissemann, submitted) we added further support to the first results of Nybom et al. (2004, 2006) that always the same chromosome sets pair during meiosis (bivalents) and the same three sets are unpaired (univalents). These findings support the hypothesis that dog roses are functional diploids: The bivalents are meiotically recombined and transmitted through the egg cell and the pollen grain, but the univalents are inherited apomic-tically by the egg cell only (Zielinski 1985, Nybom et al. 2004, Lim et al. 2005, Nybom et al. 2006). Thus, the canina meiosis combines two modes of reproduction: The interacting bivalents generate variability via sexual recombination and the apomictic univalents conserve information which allow for the tremendous variability and radiation possibilities. This unique meiosis of dog roses is presumably connected to a particular chromosome set characterized by the canina-nrITS type, which is not known to exist anymore in a diploid species (Ritz et al. 2005b). Present results show that the canina-nrITS type has at least two copies in each dog rose and is involved in bivalent formation (Kovarik et al. 2008, Ritz et al. unpublished). We assume this canina-nrITS type to be a trace for the existence of the hypothetic Protocaninae genome. The diploid Protocaninae are probably extinct but rescued as the bivalent forming set during meiosis (Fig. 2.2). On the other hand, the canina genome could also have evolved by mutation as it was shown in the evolution from teosinte to mayze in which also the existence of a “Proto-mayze” has been proposed (Doebley 2004). A polyphyletic origin of dog roses seems improbable keeping the uniqueness and complexity of canina meiosis in mind. However, phylogenetic trees based on chloroplast DNA sequences are polyphyletic with respect to the section Caninae because members of subsect. Caninae are not sister to the glandular species of subsect. Rubigineae

Fig. 2.2. Hypothetic genetic constitution of an allopolyploid pentaploid dog rose (2n=5x=35, x=7) with parental genomes from multiple hybridisation after Ritz et al. (2005a) and Kovarik et al. (2008). Black chromosomes: diploid bivalent forming Protocaninae genome, grey white and checked chromosomes: woodsii-, rugosa- and gallica- univalent forming genomes of other sections of the genus Rosa.

(19)

2. Radiation, diversity and host-parasite interaction

_____________________________________________________________________________________________________

13 H. Christ and Vestitae H. Christ but to non-dog roses of sections Indicae Thory, Rosa and

Synstylae DC. (Wissemann & Ritz 2005, Bruneau et al. 2007)

2.3 Character inheritance in the heterogamous system of dog roses

As a consequence of the canina-meiosis the proportion of genetic information contributed by the maternal parent to the offspring is four times larger than that of the pollen parent. Thus, offspring is largely matroclinal with respect to most morphological characters of leaves, flowers and hips (reviewed in Wissemann & Ritz 2007). This pattern of character inheritance is also expressed at the anatomical, biochemical and genetic level, since studies on epicuticular waxes and flower volatiles (Wissemann 2000a, Wissemann et al. 2007, Wissemann & Degenhardt, unpublished) and on microsatellites and random amplified polymorphic DNA (RAPD) bands (Werlemark 1999, Werlemark & Nybom 2001, Nybom et al. 2004, 2006, Ritz & Wissemann, submitted) demonstrated a strong matrocliny of interspecific hybrids.

However, some morphological traits do not match these observations, because Wissemann et al. (2006) showed that the growth habit of dog roses is dominantly inherited. Interspecific reciprocal hybrids between R. canina L. and R. rubiginosa L. were characterized by the lax and arching branches in contrast to dense erect branches of R. rubiginosa. This pattern could be the result of i) a heterosis effect, ii) the inheritance of dominant allele encoding growth form, or iii) which is favoured by the authors, multiple factors of which some are not inevitably subjected to inheritance but are responsible for the syndrome growth form.

Moreover, the taxonomic important characters sepal persistence and the diameter of the orifice are paternally expressed and possibly controlled by genomic imprinting (Gustafsson 1944, Ritz & Wissemann 2003). The paternal inheritance of these characters results in the morphological identity between interspecific hybrids and described dog rose species (Fig. 2.3). Ritz and Wissemann (submitted) demonstrated that R. micrantha Borrer ex Sm. and the corresponding interspecific hybrid R. rubiginosa ¯ R. canina are genetically not completely identical.

Microsatellite alleles of R. micrantha corresponded to those of the potential parents, however, all investigated samples of R. micrantha were in contrast to the pentaploid hybrids hexaploid. The authors assumed that the initial interspecific hybrid giving rise to R. micrantha was established by an increase of ploidy level to maintain two highly homologous chromosome sets for correct bivalent formation during canina meiosis.

(20)

The matrocliny of the majority of characters points despite the functional diploidy of the meiosis system to the functionality of genetic information stored on univalent genomes. Contrary, Lim et al. (2005) assumed that the lacking recombination between the univalents leads to gene degradation due to a relaxed selection pressure. However, a study on gene expression of single copy genes did not point to degradation or silencing of alleles on univalents (Ritz et al, unpublished).

Fig. 2.3. Diagram of the crossing experiment (Wissemann & Hellwig 1997) between the two dog rose species R. canina (L-type: deciduous sepals during hip ripening) and R. rubiginosa (D-type: persistent sepals during hip ripening). Both parents were used as seed and pollen parent. The arrows symbolize the direction of the crosses. Grey arrows: R. canina was used as seed parent and R. rubiginosa was used as the pollen parent. Black arrows: R. rubiginosa was used as seed parent and R. canina as pollen parent. The thickness of the arrows symbolizes the matroclinal inheritance due to the canina meiosis: The seed parent inherits 4/5 of the genome and the pollen parent only 1/5 of the genome. The outcomes of both crosses show the same type of sepal persistence as the pollen parent. The combination of the matroclinal vegetative characters (e.g. leaf surface) and the paternal type of sepal persistence is also found in already described species (R. dumalis and R. micrantha) (Reproduction from Wissemann and Ritz (2007), Fig. 2)

(21)

2. Radiation, diversity and host-parasite interaction

_____________________________________________________________________________________________________

15

2.4 Glandular trichomes matter: Rust fungi on Rosa

In Rosaceae, rust fungi have long been recognized as important parasites. Co-evolution of rusts and Rosaceae seems so strong and evident, that rust fungi can be used to determine phylogenetic relationships between certain hosts in Rosaceae (Savile 1979). El-Gazzar (1981) pointed out, that more than 1500 species from 49 genera of Rosaceae are susceptible to about 300 species out of 27 genera of the Uredinales, and that susceptibility to rust infection is strongly correlated with the chromosome base number of x=7. The most important rust fungi on Rosa are Phragmidium mucronatum and Phragmidium tuberculatum. Both have been recorded in dog roses (Gäumann 1959, Scholler 1994, Brandenburger 1994). Comparable to the findings of Evans et al. (2000) who showed the formation of subspecific strains or races of Phragmidium violaceum (Schultz) Winter, a parasite on the blackberry (Rubus L., Rosaceae), we expected the two Phragmidium species on roses to evolve and radiate in the same manner. However, ever since Savile’s publication (1979) in which the interaction between rust fungi and Rosaceae has been regarded as a co-evolutionary system, our findings weakened this assumption (Ritz et al. 2005a). Host ranges of P. mucronatum and P. tuberculatum overlapped and their infection rates did not differ between Rosa canina, R. corymbifera Borkh. and R. rubiginosa. These three species served as examples for the variation of leaf trichomes and glands observed in dog roses which might be crucial for rust infection (Bahçecioğlu & Yildiz 2005, Valkama et al. 2005). Results showed that infection by P. mucronatum and P. tuberculatum did not significantly differ between species with glabrous and hairy leaves, R. canina and R. corymbifera, respectively. However R. rubiginosa developing hairy leaves with numerous odorous glands, was significantly less infected by both species (Ritz et al. 2005a). Despite their overlapping host ranges and their morphological similarity, both fungi are genetically only distantly related:

P. mucronatum belongs to a clade of “rose rust sensu stricto” whereas P. tuberculatum is

closely related to rusts living on Rubus and Sanguisorba L. and thus explored dog roses by a host jump. The lacking host specificity of rusts on dog roses might be explained by the hybrid bridge hypothesis (Floate & Whitham 1993). It predicts that hybridisation of hosts and thus the admixture of different genomes prevents the step-by-step process of evolution and co-speciation of hosts and parasites.

2.5 Evolution and diversity of plant-pathogen-insect foodwebs on dog roses

Herbivorous insect species and their host plant species together comprise more than 50% of the macroscopic species (Strong et al. 1984). Thus, an understanding of the evolutionary driving forces as well as the ecological interactions is an important issue for our general

(22)

understanding of biodiversity. During the last 30 years the paradigm of interpreting the enigmatic diversity of insects changed from co-evolution and co-speciation to a more individualistic view (e.g. Brandl et al. 1992, Schoonhoven et al. 1998).

Macroecological work of the last twenty years showed that the insect fauna associated with a particular plant species is a complex blend of generalists and specialists (Strong et al. 1984, Tscharntke & Greiler 1995, Schoonhoven et al. 1998, Brändle & Brandl 2001). The insect fauna on a particular host depends on the available species pool of phyto-phags, the distribution and abun-dance of the host, the number of feeding niches provided by the host as well as the host’s taxonomic isolation and biochemical make-up (Strong et al. 1984, Lawton 1986, Tscharntke & Greiler 1995, Frenzel & Brandl 1998). Plant genotypes with different morphological traits may affect not only the abundance of single insect species but also the structure of associated herbivore communities (Maddox & Root 1987, Fritz & Price 1988). Whitham et al. (2003) showed that even single traits coded by few genes may have important effects on the community of exploiters (extended phenotype).

Rose bushes are characteristic features of European landscapes, in particular as components of hedges. Roses have been included into several attempts to understand α-diversity (e.g. Leather 1986). However, most of the available data on phytophages have not distinguished between the different rose species and thus ignored the bewildering diversity within that host taxon. As roses provide a fascinating example of an explosive radiation, roses are good candidates to study the effects of hybridisation as well as rapid radiation of hosts on herbivores across a flock of host plants. The available information on phytophages

Fig. 2.4. Geographical location of the 18 study sites along a transect across Germany. At all study sites the three dog rose species (Rosa canina, R. corymbifera, and R. rubiginosa) occurred together. Community study was conducted at all 18 sample sites, density analyses of Rhagoletis alternata and Diplolepis rosae at 17 study sites (all except no. 10), and the D. rosae gall community was sampled on eight sites (No. 1-3, 5-6, 9, 11, 18).

(23)

2. Radiation, diversity and host-parasite interaction

_____________________________________________________________________________________________________

17 on roses shows that there are a number of generalists attacking roses (e.g. see Zwölfer et al. 1981, Zwölfer et al. 1984), but also many specialists such as the cynipid wasp Diplolepis

rosae L., the tephritid fly Rhagoletis alternata Fallén, and the tortricid moth Notocelia roborana Dennis & Schiffermüller 1775. Ferrari for example sampled 6,000 ectophagous

insect specimens from 32 rose stands around Göttingen (see Ferrari et al. 1997). Eight out of the ten tenthredinid wasps, three out of 19 Cicadina species, and four out of 76 beetle species were specialists on roses, whereas all the 20 bug species were less specialised. For all our investigations we selected three dog rose species (Rosa canina L., R. corymbifera Borkh. and R. rubiginosa L.), all members of the dog rose section Caninae (DC.) Ser. These three species are widely distributed and abundant in central Europe and occur often in the same habitats. They are supposed to have originated by allopolyploid hybridisation events (Ritz et al. 2005b, Wissemann 2002) and expanded their range to central and northern Europe after the last ice age (Zielinski 1985). Although closely related they differ in several characters: R. canina is a glabrous rose, R. corymbifera has hairs on rhachis and abaxial leaf surface and R. rubiginosa has glandular trichomes on the lower leaf surface. Furthermore, the three rose species also differ in plant architecture (Wissemann et al. 2006) and phenology (Timmermann 1998). This leads to the question: How do these differences translate into the diversity of higher trophic levels? In the following we compare (1) the community structure of these three dog rose species, (2) the densities of two consumer specialists on the rose species and (3) the genetic adaptation of one of these specialists to the three dog rose species. Along a gradient across Germany (Fig. 2.4) we sampled at 18 study sites, where the three rose species occurred together in the same habitat.

2.5.1 Are invertebrate communities affected by leaf trichome traits of hosts?

As already noted, the European dog roses differ in a variety of morphological traits, which may influence the interactions with associated food webs. In particular, dog roses differ considerably across species in density and type of trichomes on the lower leaf surface. Trichomes are supposed to influence host choice of herbivores as well as of other invertebrates (Yencho & Tingey 1994, Zvereva et al. 1998, Ranger & Hower 2002). Although trichomes are often part of a defence system, they may, however, be beneficial for some herbivores (Eisner et al. 1998). Nevertheless, invertebrate communities may map the variation of trichomes across species (Andres & Connor 2003).

Insect communities were sampled using beating trays (diameter 70 cm, Stechmann et al. 1981) in May, June, July and August 2002 (Fricke 2004). Data were collected across 88 bushes of R. rubiginosa, 87 of R. corymbifera and 88 of R. canina (3-5 bushes per sampling site, Fig. 2.4). We sorted 75,937 individuals to insect orders (Table 2.1). Coleoptera and

(24)

Heteroptera were identified to species level to distinguish between phytophagous and non-phytophagous species. To characterise the architecture of the sampled host individuals, six variables of each bush were measured.

With a repeated measurement ANOVA we found only minor differences in the abundance of common invertebrate groups (Aphidina, Collembola, Araneae, Hymenoptera and Coleoptera) between the three rose species (Table 2.1 and 2.2). Ordinations using all major taxonomic units showed also few differences in the composition of the exploiter communities. Contrary to our expectations, the abundances of phytophagous invertebrates were higher on

R. rubiginosa than on the other two rose species. This is remarkable, because feeding

experiments showed a lower palatability of R. rubiginosa leaves using larvae of Spodoptera

littoralis (Klinge 2005; Fig. 2.5). Nevertheless, the field data suggest that most of the

phytophagous invertebrates are not negatively affected by glandular trichomes and these trichomes do not serve as a general defence against phytophages.

Apart from phytophagous invertebrates, dog roses suffer from infection by the rust fungi

Phragmidium spec. which shows significant differences in density between the three rose

species (Ritz et al. 2005a, Klinge 2005, Fig. 2.6). The species R. corymbifera and R. canina show higher infection rates than the glandular R. rubiginosa (see Chapter 2.5). By infection with these rust fungi, the lower leaf surface of the host plant is of special concern. The rust fungi infect plants by penetrating the stomata of the leaflets with the germination tubes of its uredospores. The trichomes of dog roses occur only on the lower leaf surface were also the stomata are located.

2 4 6 8 10 c onsum p tion (m g) A May B June circles = non-hybrid squares = hybrid glandular hairy glabrous 1 4 5 2 6 7 3 8 9

rose species rose species

1 4 5 2 6 7 3 8 9 C July

1 4 5 2 6 7 3 8 9

rose species

Fig. 2.5. Spodoptera leaf consumption across rose species and hybrids during three months. Corrected means ± 1 SE from a Split-Plot-Model ANCOVA (Type I) with leaf consumption as response variable in relation to the two categorical factors (month of experiment and rose genotype) and the covariables (fresh weight of larvae and specific water content). (rose species: 1 = R. rubiginosa, 2 = R. corymbifera, 3 = R. canina, 4 = R. rubiginosa x R. corymbifera, 5 = R. rubiginosa x R. canina, 6 = R. corymbifera x R. rubiginosa, 7 = R. corymbifera x R. canina, 8 = R. canina x R. rubiginosa, 9 = R. canina x R. corymbifera).

(25)

2. Radiation, diversity and host-parasite interaction

_____________________________________________________________________________________________________

19

Table 2.1. Basic information for the invertebrate groups sampled from 263 shrubs of roses during May, June, and July 2002 across 18 sites in Germany (see Fig. 2.4). Taxa were sorted by the total number of individuals (N). The first five groups (names in bold) were used for detailed analyses of the abundances. Mean: average abundance across one subsample (5 subsamples per bush). Abundance was measured as the number of individuals divided by the subsamples. Correlations of abundances (means across dates of standardised transformed abundances for each bush, n = 263) with scores of the first two principle components (PCA1 and PCA2) characterizing plant architecture. Significant correlations in bold. The principal components analysis (PCA) was performed with six measured variables of each bush. Two components passed the Kaiser criterion. The first component represented the size (PCA 1; eigenvalue: 2.31, explained variance: 38.6 %) and the second the form of an individual bush (PCA 2; eigenvalue: 1.31, explained variance: 21.8 %). PCA 1 PCA 2 Taxon N Mean s.d. r p r p Aphidina 19,656 4.97 12.44 -0.07 0.293 0.11 0.084 Collembola 19,140 4.97 8.21 -0.01 0.866 0.16 < 0.05 Araneae 11,275 2.90 3.65 -0.36 < 0.001 0.19 < 0.01 Hymenoptera 5,842 1.50 2.32 -0.23 < 0.001 0.28 < 0.001 Coleoptera 3,714 herbivorous 2,648 0.67 1.56 -0.18 < 0.01 0.02 0.759 other 1,066 0.27 0.54 -0.37 < 0.001 0.11 0.081 Thysanoptera 2,969 0.74 2.00 -0.20 < 0.001 0.18 < 0.01 Acari 2,642 0.68 1.55 0.02 0.780 0.01 0.902 Diptera 2,481 0.63 1.01 -0.22 < 0.001 0.08 0.208 Auchenorrhyncha 2,259 0.59 1.81 0.02 0.796 -0.00 0.970 Heteroptera 1,504 herbivorous 803 0.21 0.24 0.09 0.164 -0.09 0.155 carnivorous 701 0.18 0.31 0.03 0.598 0.06 0.339 Dermaptera 1,359 0.35 0.65 -0.19 < 0.01 -0.04 0.549 Psocoptera 1,147 0.29 0.58 -0.06 0.315 0.09 0.154 Lepidoptera 814 0.21 0.38 -0.06 0.340 0.02 0.775 Gastropoda 337 0.09 0.28 0.09 0.142 0.03 0.646 Planipennia 311 0.08 0.18 -0.09 0.160 0.08 0.183 Opiliones 126 0.03 0.15 -0.06 0.333 -0.06 0.363 Orthoptera 59 0.02 0.07 -0.06 0.367 -0.06 0.370 Σ 75,937

(26)

Table 2.2. Repeated measures ANOVA (type I sums of squares) for the five most abundant invertebrate groups on roses (Table 2.1). The site was a random factor, while rose species and date were fixed factors. Contrast I was defined as R. corymbifera vs. R. canina, contrast II was defined as R. rubiginosa vs. R. corymbifera and R. canina. In the part above the horizontal line, the means of each shrub over the dates were used, in the part below the line, all samples were used. Significant effects are highlighted in bold.

Aphidina Collembola Araneae Hymenoptera Coleoptera phyto.

Source of variation SS. F SS. F SS. F SS. F SS. F Foliage cover 0.33 3.0 5.99 54.3 3.22 86.3 1.54 38.3 0.04 1.8 PCA 1 0.12 1.1 0.29 2.6 2.37 63.5 0.50 12.4 0.39 16.0 PCA 2 0.42 3.9 0.13 1.2 0.29 7.7 0.76 18.9 0.00 0.1 Site 14.35 7.8 26.20 14.0 12.72 20.1 4.56 6.7 4.87 11.9 Rose species 4.69 7.1 0.23 0.4 0.21 1.7 0.17 1.0 0.00 0.0 Contrast I 0.11 0.6 0.25 1.2 0.00 0.0 0.01 0.4 0.00 0.0 Contrast II 4.60 9.9 0.01 0.0 0.21 3.4 0.16 1.2 0.00 0.0 Site x rose species 11.25 3.0 10.17 2.7 2.15 1.7 2.93 2.1 2.28 2.8 Site x contrast I 3.24 1.8 3.71 2.1 1.06 1.7 0.66 1.0 0.52 1.6 Site x contrast II 7.92 4.0 6.19 3.0 1.04 1.6 2.29 3.4 1.76 4.2

Residuals I 22.42 22.72 7.68 8.29 4.96

Residuals I (contr I) 14.20 14.49 5.09 5.28 2.59 Residuals I (contr II) 25.85 26.92 8.79 8.95 5.48

Date 14.74 11.1 31.18 22.2 21.83 87.7 5.68 14.4 4.36 17.8 Date x foliage cover 2.21 15.1 1.44 12.8 0.20 5.1 1.71 35.0 0.67 20.8 Date x PCA 1 0.17 1.2 0.03 0.3 0.04 0.9 0.74 15.1 0.10 3.0 Date x PCA 2 0.02 0.2 0.36 3.2 0.22 5.7 0.14 2.8 0.12 3.8 Date x site 22.54 9.0 23.90 12.5 4.23 6.3 6.72 8.1 4.17 7.6 Date x rose species 2.95 5.2 0.32 0.6 0.44 4.8 1.01 4.8 0.65 5.1 Date x contrast I 0.22 1.5 0.00 0.0 0.07 2.1 0.08 1.4 0.06 1.8 Date x contrast II 2.75 6.5 0.31 1.3 0.37 6.5 0.95 6.0 0.59 6.2 Date x site x species 9.65 1.9 8.57 2.2 1.56 1.2 3.62 2.2 2.15 2.0 Date x site x contr I 2.54 1.0 4.38 2.2 0.58 0.9 0.95 1.2 0.56 1.2 Date x site x contr II 7.17 2.9 4.17 2.0 0.96 1.4 2.67 3.2 1.60 2.9

Residuals II 30.27 23.23 8.19 10.09 6.63

Residuals II (contr I) 21.08 15.93 5.13 6.39 3.82

Residuals II (contr II) 32.95 27.64 8.87 11.10 7.25

(27)

2. Radiation, diversity and host-parasite interaction

_____________________________________________________________________________________________________

21 Thus, the trichomes might serve as a defence against rust infections. This idea is not new and was already proposed for other plant species by Bahçecioğlu & Yildiz (2005) and Valkama et al. (2005). Furthermore, in the glandular trichomes of R. rubiginosa secondary

compounds belonging to the sesquiterpenes occur (Klinge, unpublished). These substances are known to inhibit the growth of fungi (e.g. Alvarez-Castellanos et al. 2001, Cakir et al. 2004). Overall our data do not support the hypothesis that trichomes evolved as a defence strategy against inverte-brates and especially against herbivores. How-ever, our data are consistent with the hypotheses that trichomes are a defence strategy against rust fungi.

2.5.2 Do Rhagoletis alternata and Diplolepis rosae differ in density between the three rose species?

Two highly specialised consumer species of dog roses are the European rose-hip fly

Rhagoletis alternata Fall. (Diptera, Tephritidae) and the rose gall wasp Diplolepis rosae L.

(Hymenoptera, Cynipoidea). The fruit fly Rh. alternata infests the fleshy fruits of species from several Rosa sections (White 1988). Adults emerge in early summer and females oviposit into green hips marking the attacked hips by an oviposition-deterring pheromone (Bauer 1986). Larvae feed exclusively in the hypanthium and do not attack the seeds. Mature larvae leave the hips for pupation and hibernate in the soil (Bush 1992). The percentage of infested hips per shrub is usually high, frequently reaching 100% (Bauer 1998). Overall, Rh. alternata has little impact on the sexual reproduction of roses and thus on the fitness of the hosts (Bauer 1998).

Fig. 2.6. Density of Phragmidium spp. across the three dog rose species (grey circle = Rosa rubiginosa, dark-grey square = R. corymbifera, black triangle = R. canina) and the geographical locations in the year 2002. Corrected means were calculated with General Linear Model analyses. Geographical location = ranking of the 18 sites from South to North, from West to East and from high to low altitudes based on results from a principal component analysis.

(28)

The holarctic cynipid wasp D. rosae is a univoltine gall maker (Adler 1877). In Europe, the conspicuous and multichambered galls have been found on Rosa species from several sections (Schröder 1967). In southern Sweden Stille (1984) found all species on which

D. rosae occur belonging to the section Caninae. Due to the physiological manipulation of

the host plant, cynipid gall wasps are closely adapted to their host plants (e.g. Crawley & Long 1995, Kato & Hijii 1997). The relationship between plants and gall inducing insects are usually very specific, suggesting tight co-evolutionary processes (Hilker et al. 2002). Galls develop as a result of interactions between the inducing insect and plant, wherein the insect gain control and redirect the growth and physiology of attacked organs to the insects’ advantage (Shorthouse et al. 2005b). Both consumer species are known to attack all the three rose species. However, do the consumer species show differences in density between the host species? Klinge (2005) monitored the density of larvae of Rh. alternata and D. rosae

Fig. 2.7. Patterns of density across the three dog rose species and the geographical locations in the two study years 2002 and 2003 for (A, B) Rhagoletis alternata and (C, D) Diplolepis rosae (corrected means were calculated with General Linear Model analyses). Geographical location = ranking of the 18 sites from South to North, from West to East and from high to low altitudes based on results from a principal component analysis.

(29)

2. Radiation, diversity and host-parasite interaction

_____________________________________________________________________________________________________

23 galls for each rose species in September 2002 and 2003 on 3 to 5 randomly selected bushes within each sample site (Fig. 2.4). To monitor the density of Rh. alternata 50 hips from each bush were collected haphazardly and the percentage of infested hips was used as an estimate of density. During the same sampling dates, rose bushes were searched for galls of

D. rosae. The total number of galls on each bush was used as a measure of gall density.

Densities of the two consumer species varied between host plant species and geographical location (Fig. 2.7, Table 2.3). The highest densities of the two phytophages were found on the odorant R. rubiginosa. Although the density of Rh. alternata and D. rosae exhibited significant variations between sites, we found no general geographical trends. Two suites of factors may account for these differences: ecological and genetic factors. Plant phenology (Hodkinson 1997) and abundance of natural enemies (Koptur 1985) may be such possible factors or genetic variation of the consumer species may also trigger variation in density. On the sexual reproduction of the roses Rh. alternata had little impact. Even when larvae attacked all hips of a bush, we did not find a negative impact of Rh. alternata densities on reproduction (number of hips) and leaf cover in the following year. Furthermore, Rh. alternata density patterns across years were highly predictable: Host plants with a high attack in one

year showed also a high attack in the following year (r2 = 0.45, P < 0.05, N = 212). We

interpret the predictable fruit production as further evidence that larvae of Rh. alternata have overall little impact on the fitness of roses.

In contrast to Rh. alternata, the gall-forming D. rosae manipulates the physiology of the rose bushes to produce the gall and gall tissue (Bronner 1992, Bayer 1992, Bagatto et al. 1996, Harper et al. 2004). Although significant top-down effects of gall wasps on the population dynamics of the hosts seem to be rare (e.g. Stone et al. 2002), there is evidence that high cynipid densities can negatively affect host plant growth (e.g. Crawley & Long 1995, Kato & Hijii 1997). Due to the low density of D. rosae galls, with a mean density of 0.17 galls per bush, the roses did not show a negative response: The density of hips as well as leaf cover

was independent of the number of galls in the previous year (hip density: r2 = 0.02, P = n.s.,

leaf cover: r2 = 0.02, P = n.s.).

2.5.3 Does Rhagoletis alternata form host races on the three dog rose species?

Phytophagous insects may adapt to host plants, thereby forming host races as a first step during sympatric speciation. Host races are sympatric but genetically differentiated populations of exploiters that use different host species (Drès & Mallet 2002).

(30)

Table 2.3. Repeated measures ANOVA (type I sums of squares) for the densities of Rhagoletis alternata and Diplolepis rosae. The site was a random factor, while host species (Rosa canina, R. corymbifera and R. rubiginosa) and year were fixed factors. Differences in shrub characteristics were accounted for with the two principal components (PC1 and PC2) derived in a PCA. PC1 characterizes the size of a shrub (35% explained variance; variables with loading > 0.6: height, diameter, diameter of the largest shoot) and PC2 the foliar density of a shrub (26% explained variance; loadings > 0.6: density of leaflets and hips, leaf cover). SS = sum of square, df = degrees of freedom; the F-ratio was calculated with the appropriate error term; significant effects are marked in bold.

Rh. alternata D. rosae SS df F SS df F PC1 0.16 1 2.24 0.59 1 1.24 PC2 0.20 1 2.71 16.93 1 35.61 Site (S) 38.73 16 33.63 34.33 16 4.51 Host (H) 1.15 2 3.69 12.86 2 6.01 S x H 4.98 32 2.16 34.21 32 2.25 Residuals 11.45 159 82.26 173 Year (Y) 10.77 1 48.09 10.11 1 18.14 Y x PC1 0.18 1 5.99 0.02 1 0.06 Y x PC2 0.18 1 6.21 4.61 1 17.45 Y x S 3.58 16 7.61 8.92 16 2.11 Y x H 0.29 2 2.01 2.54 2 1.89 Y x S x H 2.28 32 2.42 21.53 32 2.55 Residuals 4.68 159 45.65 173

Vaupel et al. (2007) studied the population genetic structure of Rh. alternata along the geographic gradient and in relation to the three rose species. They collected larvae from 15 sites across Germany and from three valleys of Valais in Switzerland. They were able to score nine allozyme loci (five polymorphic). Populations from the three hosts did not differ in genetic variability. These results provide two further unexpected findings. Firstly, although they found significant genetic differentiation between populations from different host species, the differentiation was very low (0.9%) and cannot be interpreted as an indication for host races. A reason may be the permanent and ongoing hybridisation between rose species of the section Caninae. Secondly, they found surprisingly little geographic structure of genetic differentiation between populations of this fruit fly across central Europe. Additionally, analysis of amplified and sequenced fragments of the mitochondrial genes encoding cytochrome oxidase I (800 bp), cytochrome oxidase II (470 bp), and cytochrome b (450 bp), indicated that all individuals of Rh. alternata (n = 21) from several sites in Europe shared the

(31)

2. Radiation, diversity and host-parasite interaction

_____________________________________________________________________________________________________

25 same haplotype (Kohnen et al. 2009, see Chapter 6). This lack of genetic variation is unexpectedly low compared to data of other insect taxa (p = 0.0016, n= 63).

Three mutually non-exclusive reasons may explain these findings. Firstly, gene flow between populations of Rh. alternata is high. Secondly, the pattern of genetic differentiation is based on a recent expansion of the distributional range or a host shift of the fly. Thirdly, symbionts, such as Wolbachia, shape at least mtDNA evolution (Hurst & Jiggins 2005). During the initial phase of symbiont invasion, selective sweeps may reduce mtDNA diversity, thereby producing a genetic signal similar to that produced by a population bottleneck with subsequent expansion (Hurst & Jiggins 2005).

Because of the low gene flow estimated from allozyme data Rh. alternata seems to be a good disperser (Leclaire & Brandl 1994, Vaupel et al. 2007). Even the Alps do not seem to be a geographical barrier for gene flow between populations. In part this is explained by the behaviour of the females which mark the hips with a pheromone after oviposition (Bauer 1986, 1998). Often a high proportion of hips (up to 100%) are infested. Females leave such localities and search for rose shrubs with a lower proportion of infested hips. Vaupel et al.

(2007) found no isolation by distance (r2 = 0.01, p = 0.19), which may indicate that

populations are not in an equilibrium between drift and gene flow (Hutchison & Tempelton 1999). The time to reach population genetic equilibrium increases with population size. The population size of tephritids is often very high (McPheron et al. 1988), and Rh. alternata is no exception. Low genetic differentiation and lack of isolation by distance points to a recent range expansion (Hutchison & Templeton 1999). Such an expansion could be induced by colonisation events after the last ice ages or by very recent host-shift events. In both cases, a low number of founder individuals would lead to a population bottleneck. Lower levels of genetic variation would be expected (Harrison 1991). Rh. alternata is a specialist on members of the genus Rosa section Caninae and therefore dependent on the distribution of its host. These dog roses originated by hybridisation events during the last ice ages (Ritz et al. 2005b, Wissemann 2002) and re-colonised Europe afterwards (Dingler 1907). Founder individuals of the fly species may have shifted to this new host, providing an explanation for the low genetic variability. Overall the considerable gene flow between populations of

Rh. alternata, limited phenological differences between host species, and the ongoing

hybridisation of hosts may prevent the formation of genetic differences between populations of exploiters on rose species.

2.6 How are the differences between the three closely related dog rose species

translated into higher trophic levels?

Due to the enclosed environment within the galls, gall-makers and their parasitoids provide a good opportunity to analyse tritrophic interactions. Gall characteristics like shape and

(32)

toughness are plant derived structures, but often regulated by insect genes whereas the gall diameter for example is regulated by plant genotype (in Salix lasiolepis; Price & Clancy 1986b). In turn the galls sizes and densities determine the success as well as composition of parasitoid communities (Brandl & Vidal 1987, Schlumprecht 1989, Weis 1983).

So far, we showed certain differences in community structure and consumer density between the glandular rose R. rubiginosa and the other two rose species. The galls of D. rosae form the basis of a complex community of an inquiline and at least 12 species of parasitoids (Blair 1944, Redfern & Askew 1992). Beside D. rosae, the inquiline Periclistus brandtii R. (Hym. Cynipidae) and several parasitoid wasp species can be found within the galls (Redfern & Askew 1992). The gall-maker D. rosae is parasitized by at least five parasitoid species:

Orthopelma mediator Thunb. (Hym. Ichneumonidae), Torymus bedeguaris L. (Hym.

Torymidae), Pteromalus bedeguaris Thomson (Hym. Pteromalidae), Glyphomerus stigma Fabr. (Hym. Torymidae) and Eupelmus urozonus Dalman (Hym. Eupelmidae). The dominant one which is almost invariably present is O. mediator (Stille 1984). The inquilin P. brandtii utilizes the galls to create its own chambers on the surface of the gall. The effect of P.

brandtii attack on the gall is so far unknown, either it enlarges the gall or reduces the space

otherwise available to D. rosae. The inquilin is also parasitized by G. stigma and

E. urozonus, but additionally by Caenacis inflexa Ratzeburg (Hym. Pteromalidae) and Eurytoma rosae Nees (Hym. Eurytomidae).

We sampled all available D. rosae galls at eight of the sample sites (Fig. 2.4) and hatched the galls outside until all inhabitants emerged (Table 2.4). With a generalised linear model the parasitism rate of D. rosae varied between sites as well as rose species and decreased with increasing gall volume (Klinge 2005, see Chapter 3). Whereas the mean number of

D. rosae galls was highest on R. rubiginosa, the mean parasitism rate was lowest on this

rose species. But all two way interactions between host species and site were also significant, pointing to complex effects of geography and rose species on the communities

associated with D.rosae galls. As mentioned above, R. rubiginosa is the only rose species

we examined with glandular trichomes rich in secondary metabolites as sesquiterpenes. Host plant variation and differences in secondary metabolites often influence higher order interactions within insect communities (Eisenbach 1996, Fritz et al. 1997, Gange 1995, Marquis & Whelan 1996, Prezler & Boecklen 1994).

Beyond factors intrinsic to the host plant, local environmental conditions have also some influence on plant traits e.g. resulting in different nutritive quality and thereby changing interactions with herbivores and their natural enemies (Moon & Stiling 2000, Price & Clancy 1986b, Stiling & Rossi 1997). Even effects of genetic differences among and within host plant taxa may be modified by environmental conditions (Fritz et al. 1997).

Referenzen

ÄHNLICHE DOKUMENTE

Using natural isolates of parasites to determine specificity can greatly underestimate specificity in host-parasite interactions. ramosa, we find much higher

I established that the infection of Daphnia magna by Pasteuria ramosa could be decomposed in at least five sequential steps (Chapter 1): 1) the encounter between the host and

fimbriatus by its larger size (snout-vent length up to 200 mm vs. 295 mm), hemipenis morphology, colouration of iris, head and back, and strong genetic differentiation (4.8 %

ILCI (Indian Languages Corpora Initiative) started by Technology Development for Indian Languages (TDIL) program of Ministry of Communication and Information

Joint statement by PNND Co-Presidents Uta Zapf MdB (Germany), Hon Marian Hobbs MP (New Zealand), Senator Abacca Anjain Maddison (Marshal Islands), Alexa McDonough MP (Canada)

[10] This approach revealed one of the major challenges of expanding the genetic code: the evolution of additional, noninteracting (orthogonal) translational

The expansion or adopted model of maritime and linear and geometric Bell Beaker pottery almost certainly began in this period, as well as the associated rites, given the antiquity

My results show, first, that less severe cyclical fluctuations for both series are observed over time and, second, a weakening relationship of these cyclical fluctuations between